Lipopolysaccharide from the commensal microbiota of the breast enhances cancer growth: role of S100A7 and TLR4

Abstract

The role of commensal bacterial microbiota in the pathogenesis of human malignancies has been a research field of incomparable progress in recent years. Although breast tissue is commonly assumed to be sterile, recent studies suggest that human breast tissue may contain a bacterial microbiota. In this study, we used an immune-competent orthotopic breast cancer mouse model to explore the existence of a unique and independent bacterial microbiota in breast tumors. We observed some similarities in breast cancer microbiota with skin; however, breast tumor microbiota was mainly enriched with Gram-negative bacteria, serving as a primary source of lipopolysaccharide (LPS). In addition, dextran sulfate sodium (DSS) treatment in late-stage tumor lesions increased LPS levels in the breast tissue environment. We also discovered an increased expression of S100A7 and low level of TLR4 in late-stage tumors with or without DSS as compared to early-stage tumor lesions. The treatment of breast cancer cells with LPS increased the expression of S100A7 in breast cancer cells in vitro. Furthermore, S100A7 overexpression downregulated TLR4 and upregulated RAGE expression in breast cancer cells. Analysis of human breast cancer samples also highlighted the inverse correlation between S100A7 and TLR4 expression. Overall, these findings suggest that the commensal microbiota of breast tissue may enhance breast tumor burden through a novel LPS/S100A7/TLR4/RAGE signaling axis.

Keywords: LPS; RAGE; S100A7; TLR4; breast cancer; microbiota.

Fig. 1

Comparative phenotypes of microbiome samples. Scatter plots showing the relative abundance of predicted phenotypes of different groups found in EST (early stage), ESTSK (early‐stage skin), LST (late stage), LSTDSS (late‐stage dextran sodium sulfate‐induced), and LSTSK (late‐stage skin) breast tumor samples. Predicted phenotypes include (A) aerobic, (B) anaerobic, (C) facultative anaerobic, (D) biofilm‐making, (E) Gram‐positive, (F) Gram‐negative, (G) microbiota that is potentially pathogenic, and (H) stress‐tolerant ability using the bugbase taxonomy prediction software. The gray horizontal line in each graph represents median values derived from the three biological replicates of each of the EST, LSTDSS, and ESTSK groups, four biological replicates of each of the LST and LSTSK groups, and two biological replicates of the control group (NMG). *indicates significant (P < 0.05) up or down relative abundance as compared to the control group as indicated in the figure and evaluated based on the Kruskal–Wallis and Mann–Whitney U‐test as described in Materials and methods.

Fig. 2

The relative abundance of bacteria at phylum level and genus level corresponding to different samples. (A) The Sankey plots showing the relative abundance of bacteria at phylum level (middle) and genus level (right) corresponding to different samples (left), visually displaying the species annotation information, corresponding relationship, and proportion of the two levels. The plot shows flow changes of the data, and the width of the plots indicates the size of the flow. (B) Average taxonomy community abundance in a different group of samples. The horizontal axis in the figure is the name of the sample, while the vertical axis represents the relative abundance of a certain classification. Different colors correspond to different phylum indicate the composition and expression of species within and between groups. According to the sample species abundance table, the 30 species with the highest abundance were selected by default for classification or functional classification. The plus (+) sign in different columns and at the right side next to the list of phylum indicates Gram‐positive organisms. (C) The heat map showing the mean value within the group is selected to represent the relative abundance of the group. Blue indicates lower abundance, and red indicates higher abundance. The right‐side list is 31 major significant genera identified in a different group (horizontal axis). The + (plus) sign indicates Gram‐positive bacteria. The rest of the genera are Gram‐negative. The gradient scale (extreme right side) from blue to red in the heat map reflects the change of abundance from low to high. The closer to blue and red indicates the lower the abundance and the higher the abundance, respectively. The results are derived from three biological replicates of each of the EST, LSTDSS, and ESTSK groups, four biological replicates of each of the LST and LSTSK groups, and two biological replicates of the control group (NMG).

Fig. 3

S100A7 expression inversely associated with TLR4 expression in breast tumor. (A) Bar diagram representing the level of LPS (endotoxin) in the mammary fat pad of each group. The data presented here are the mean ± SEM of three biological triplicate (n = 3). One‐way ANOVA (Tukey's multiple comparisons test) was used to calculate the P‐values. (*P < 0.05, **P < 0.01, and ***P < 0.001, ns indicates nonsignificant). (B) Expression of S100A7 and TLR4 was analyzed in tissue lysates harvested from each group by western blot. β‐Actin (ACTB) was used as a loading control. Bar diagrams represent the fold changes of S100A7 and TLR4 in three biological replicates of all the different groups. The data represented the mean ± SEM of three biological replicates (*P < 0.05, **P < 0.01, and ***P < 0.001, ns indicates nonsignificant). One‐way ANOVA (Tukey's multiple comparisons test) was used to calculate the P‐values. (C) Representative immunohistochemistry (IHC) images of same malignant breast tumor tissues stained with S100A7 and TLR4 antibodies. Scale bar = 300 μm. (D) Spearman’s correlation analysis of S100A7 and TLR4 protein expression in tissue microarrays (TMAs) containing malignant breast tumor tissues (n = 20) and their normal adjacent controls (n = 20). (E) Spearman’s correlation analysis of S100A7 and TLR4 protein expression only in malignant breast tumor tissues (n = 20). Two‐tailed t‐test was used to calculate the P‐values.

Fig. 4

High expression of S100A7 and low level of TLR4 correlate with poor prognosis of invasive breast cancer patients. Expression of (A) S100A7 and (B) TLR4 was analyzed in normal breast tissues and different intrinsic subtypes of breast cancer by using TISIDB database. Normal (n = 137), LumA (n = 508), LumB (n = 191), and basal (n = 172). The Kaplan–Meier (KM) plotter analysis of (C) S100A7 (PSOR1) and (D) TLR4 in basal subtype of breast cancer patients. The log‐rank test was used for statistical comparison of two groups. Representative immunohistochemistry (IHC) images of (E) S100A7 and (F) TLR4 staining in malignant breast tumors (n = 29) and their adjacent normal controls (n = 23). Scale bar = 300 μm. Graphs representing the percentage (%) of high S100A7‐ or TLR4‐positive cells in malignant breast tumors (n = 29) and their adjacent normal controls (n = 23). (**P < 0.01, ***P < 0.001). A nonparametric Kolmogorov–Smirnov (KS) test was applied to calculate the P‐value.

Fig. 5

Effect of LPS treatment on the expression of S100A7 in breast cancer cells (A) SUM159 cells were treated with 10 and 100 ng·mL⁻¹ of LPS or PBS (0) for 24 h, and the protein expression of S100A7 and TLR4 in SUM159 cells was analyzed by western blot analysis. GAPDH was used as a loading control. (B) Immunofluorescence (IF) analysis of S100A7 expression in PBS and LPS (100 ng·mL⁻¹)‐treated SUM159 breast cancer cells. (C) The levels of S100A7 mRNA transcripts were determined in SUM159 cells after treatment with PBS or 100 ng·mL⁻¹ LPS for 24 h by qRT‐PCR analysis. GAPDH was used as a loading control. The data presented here are the mean ± SEM of three biological triplicate (n = 3). (D) SUM159 cells after pretreatment with LPS blocker, polymyxin B (PMB), at a final concentration of 30 μg·mL⁻¹ for 1 h were treated with 100 ng·mL⁻¹ LPS for 24 h, cell lysates were collected, and S100A7 expression was analyzed by western blot. GAPDH was used as a loading control. (E) Cell viability analysis of SUM159 treated with PMB in the presence or absence of LPS after 24 h. The data presented here are the mean ± SEM of triplicate experiments (n = 3). (F) Expression of S100A7 and TLR4 was analyzed in MDA‐MB‐468 cells treated with different concentrations of LPS for 24 h. GAPDH was used as a loading control. (*P < 0.05, **P < 0.01).

Fig. 6

LPS‐mediated activation of S100A7 counteracts TLR4 expression in breast cancer cells. Expression of (A) S100A7 and (B) TLR4 proteins was analyzed in MDA‐MB‐231 vector (231V)‐ and S100A7‐overexpressing MDA‐MB‐231 (S7OE) breast cancer cells by western blot analysis. (C) Expression of S100A7 was analyzed in S7OE breast cancer cells treated with LPS either in the presence or in absence of PMB for 24 h. (D) Expression of TLR4 in 231V and S7OE breast cancer cells was analyzed after treatment with PBS or LPS for 24 h by western blot. GAPDH was used as a loading control. Effect of LPS on (E) wound healing and (F) migrating abilities of 231V and S7OE breast cancer cells. The data presented here are the mean ± SEM of triplicate experiments (n = 3). One‐way ANOVA (Tukey's multiple comparisons test) was used to calculate the P‐values (*P < 0.05, **P < 0.01).